Nodes

What a node is#

A node is one machine in a DevLyft cluster, in a specific location. Each replica of your deploy runs as a pod on a node.

• Shared compute nodes are pooled. Your pods run next to other organisations' pods.
• Isolated compute places your pods on dedicated nodes, whole machines of a fixed machine type used only by your organisation.

The Nodes page#

Open Infrastructure→Nodes. The table lists nodes across all of your projects. It refreshes from the Refresh control at the top of the page.

Node

The node's name.

Location

The location ID the node runs in, the same ID you pick in a deploy's Location field.

Arch

CPU architecture of the machine. Your image needs a variant for this architecture.

Status

Running when the node is ready and accepting new pods. Stopped in any other case. See Node health.

CPU

The node's total vCPU.

Memory

The node's total memory.

IP

The node's IP address.

Each row has a ⋮ menu with View metrics. Node metrics only exist for dedicated nodes (see Node metrics).

Machine types#

A dedicated node's size comes from its Machine type, chosen on a dedicated environment or deploy. The list comes from the platform, so new types appear without a dashboard change. Each type shows a name and spec, for example Balanced 2 — 2 vCPU · …. Today's sizes are 2 and 4 vCPU. A machine type must be offered by a cluster in the deploy's location.

Node metrics#

View metrics on a node shows CPU, Memory, Net In and Net Out for the whole machine. These totals are only available for dedicated nodes. A shared node's totals would mix other tenants' usage, so the API refuses them. For shared workloads, use deploy or pod metrics instead. See Metrics.

"Node data is currently unavailable"#

If the Nodes page shows this panel, node listings couldn't be loaded for your login. Your deploys aren't affected. Check back later, or use the Pods page and deploy metrics in the meantime.